A map of the processing rules used across GeoLira: source registration, freshness, canonical records, quality gates, derived calculations and fail-closed behavior when evidence is missing.
Data chain from source to screen
GeoLira does not present a raw provider response as finished truth. Data pass through source registration, parsing, a canonical record, unit and quality checks, freshness policy and only then the API and user interface. Each record preserves provenance, observation time and source identity so the displayed result can be audited.
Measurement, derivation, model and authored index
The science layer distinguishes direct measurements, derived analysis, model products and authored indices. Kp, Bz and solar-wind speed are scientific data with defined providers. Lunar illumination is a local astronomical derivation. Earth Pulse and Earth Vibration are versioned 0-100 authored indices and are not physical units.
Quality, fail-closed behavior and inference boundaries
Missing data are never silently converted to zero. A record may be rejected for bad quality, invalid range, station failure or excessive age. When evidence is insufficient the result is PARTIAL or UNAVAILABLE instead of falsely precise. Geophysical data are not used to infer medical diagnosis or personal causation, and scientific, community, personal and symbolic layers remain separated.